(1) The energy savings associated with the roosting behaviour of barn owls (Tyto alba) were determined with a biophysical model using measurements of microclimate from a roost and nest site in SW Scotland (55°10' N 3°12' W) from April 1991'March 1992. (2) The roost building provided complete shelter from wind and precipitation. Air temperature inside the roost building was 1.4°C greater than ambient and matched the seasonal change in temperature. Air temperature inside the nest box was on average only 0.8°C greater than ambient but was 2'3°C warmer when adults and chicks were in the nest during the breeding season. (3) Estimated metabolic heat production was significantly different between locations and averaged 67.9, 68.1, 75.5 and 84.2 W m'2 for a barn owl roosting in the building, nest box, spruce tree and in the open, respectively. At night metabolic heat production was greater by 4'12% compared with daytime, depending on location. (4) Heat loss was 30% greater in winter months than in the summer in all locations. By roosting in the building an owl would make savings of 21.6 W m'2 in March but only 12.9 W m'2 in August. In a tree roost a barn owl would save 11.8 W m'2 in March and 5.8 W m'2 in August. (5) Barn owls were estimated to reduce metabolic heat production by 19% by roosting in the building and by 10% by roosting in a tree. In the building and tree savings of 21 and 9% occurred during the day compared with 17 and 12% at night. (6) Metabolic savings were strongly dependent on weather conditions with average metabolic savings of 26% occurring in wet and windy conditions compared with only 12% on dry-calm days. Maximum savings of 29'36% occurred on wet days. (7) Barn owls appear to compensate for high metabolic demands for heat production by taking advantage of better thermal conditions within buildings, especially during the day when metabolic savings are greatest.